iiinova — where your digital self keeps evolving

Nutan Limaye @nutanlimaye.bsky.social · 07/10/2026

Very likely not the end. The quality of the papers has improved. The related work section is richer in several papers.

100

Nutan Limaye @nutanlimaye.bsky.social · 06/10/2026

It is interesting that the paper does not have authors. The writers of the paper mention themselves in acronyms and mention the engineer who obtained the result through prompts in the first few paragraphs. Seems like the paper resolves an important open problem. Does anyone understand the result?

Nutan Limaye @nutanlimaye.bsky.social · 06/10/2026

What would you do if you were contacted by a company in this way? What "should" one do is a different question.

Nutan Limaye @nutanlimaye.bsky.social · 06/10/2026

The way it happened: The internal model of Claude developed the algorithm. "Anthropic shared the algorithm with the authors in September 2026 under a confidentiality agreement, offered compensation, and provided access to the public version of Claude."

150

Nutan Limaye @nutanlimaye.bsky.social · 28/09/2026

hahahha! Meant to say AI agents, but agree with the AI gents interpretation. :D :D

Nutan Limaye @nutanlimaye.bsky.social · 23/09/2026

OpenAI has solved 100-odd long-standing open problems. openai.com/index/adviso... Now they are wondering how to release them. The best mind of today are now deciding on policy of release rather than thinking about the problems and solutions. I never saw this coming. agmai.org

openai.com

Advisory Group on Mathematics and Artificial Intelligence

OpenAI is working with an independent Advisory Group on Mathematics and Artificial Intelligence to guide the review and communication of emerging AI results.

100

Nutan Limaye @nutanlimaye.bsky.social · 17/09/2026

Just looked up something out of curiosity. 2023, 2024, 2025: number of papers on ECCC till around mid-September was around 140. 2026: the number is 192.

Nutan Limaye @nutanlimaye.bsky.social · 16/09/2026

We sold mathematics by talking about its big open problems, but perhaps, the correct narrative should have been, about what we will understand when we solve them.

Nutan Limaye @nutanlimaye.bsky.social · 16/09/2026

Oded Goldreich wrote this in 2025, but this resonates with me. |https://www.wisdom.weizmann.ac.il/~oded/on-proofs-t2.html What I agree with: Achievements are a proxy for understanding. Achievements without understanding do not matter. My take: +

141

Nutan Limaye @nutanlimaye.bsky.social · 14/09/2026

This is not where I expected TCS to be in 2026! The last line in the abstract is distressing. :( Is this how we need to work when we get a big result? As if someone is always chasing us?

0422

Reposted by Nutan Limaye

Ali @alisouthhams.bsky.social · 12/08/2026

Courtesy of Hats on Cats 🤣 #solareclipse

25577371417

Nutan Limaye @nutanlimaye.bsky.social · 13/08/2026

"Even a broken clock is right twice a day". An idiom that will be lost on the next generation! :D

static.klipy.com

Pressing Button on Wooden Digital Clock

ALT: Pressing Button on Wooden Digital Clock

Clément Canonne @ccanonne.github.io · 01/08/2026

Reposted by Nutan Limaye

- be there for the junior researchers, and be generous with your funding/opportunities. Again, You (we? I feel old) have more stability, more of a cushion. Go out of your way to ease the path of the next generation and shield them. 4/ (end of personal, biased advice for senior TCS researchers)

1402

Nutan Limaye @nutanlimaye.bsky.social · 29/07/2026

The WAVE (Women in Algorithms, Venture into Exploration) 2026 workshop will take place on Friday, 11 September 2026, at the University of Copenhagen, Denmark. Website: wave-workshop.github.io Registration free but mandatory. Register now!! Please spread the word. Write to us if you have queries.

072

Nutan Limaye @nutanlimaye.bsky.social · 20/07/2026

Thore Husfeldt and I wrote a guest post on @lance.fortnow.com and Bill Gasarch’s blog trying to explain the recent result that bipartite perfect matching allows *deterministic* efficient parallel algorithms (“BPM belong to NC”). blog.computationalcomplexity.org/2026/07/bipa...

blog.computationalcomplexity.org

Bipartite Perfect Matching in Deterministic NC

Nutan Limaye and Thore Husfeldt guest post on the new deterministic parallel algorithm for bipartite perfect matching by Abhranil Chatterje...

081

Nutan Limaye @nutanlimaye.bsky.social · 05/07/2026

It took more than 24 hrs. Close to 26 hrs.

110

Nutan Limaye @nutanlimaye.bsky.social · 04/07/2026

Crossing the English Channel in this beautiful ferry.

040

Nutan Limaye @nutanlimaye.bsky.social · 03/07/2026

Heading to ICALP 2026. From Copenhagen to London by bus!!! 🚌

390

Pro tip for listening: we are much more fun and snappy at 1.5x or 2x speed. So you are highly encouraged to speed up the audio. 🙂

This is also the final episode of Season 1 of Life of a Researcher. Interestingly, this episode is a collaboration with dear Thore Husfeldt, who also has a podcast of his own (CAST-IT)! The intro to this episode is a little different from usual because of the collaboration.

100

Recipient of both the Abel Prize and the Turing Award, Wigderson has been instrumental in revealing deep connections between computation, mathematics, and the nature of efficient reasoning.

100

Our guest is none other than Avi Wigderson! He is one of the leading figures in the theory of computation. His groundbreaking work on computational complexity, randomness, and cryptography has helped define the field over the past four decades. +

110

Nutan Limaye @nutanlimaye.bsky.social · 15/06/2026

Here is yet another episode of Life of a researcher. open.spotify.com/episode/4M3T... This episode is really special. The picture says it all, I suppose. But read the comments to know more! Pro tip for listening: we are much more fun and snappy at 1.5x or 2x speed. So speed up the audio!!

180

Now it's official!! I already finished reading the introduction. Working out the proof is what I'm looking forward to next! Highest on my to-do. eccc.weizmann.ac.il/report/2026/... Paper number 100. Love it!! :D

eccc.weizmann.ac.il

ECCC - TR26-100

050

Reposted by Nutan Limaye

Joshua Grochow @joshuagrochow.bsky.social · 14/06/2026

The preprint is out! Bipartite Perfect Matching is in NC, by Abhranil Chatterjee, Sumanta Ghosh, Rohit Gurjar, Roshan Raj, Thomas Thierauf eccc.weizmann.ac.il/report/2026/... #MathSky

eccc.weizmann.ac.il

ECCC - TR26-100

2265

Nutan Limaye @nutanlimaye.bsky.social · 11/06/2026

If you know it you know it. www.kva.se/en/event/cel...

We are ending with a lovely talk by Tuukka Korhonen who is telling us about lower bounds on tropical circuits, circuits that work over the (max, +) semiring.

Pietro Posta is telling us how to count interesting representation theoretical quantities using quantum algorithms. Specifically, he provides #BQP algorithms for counting these quantities.

The first talk in the final session of the final day of the workshop Julian Dörfler is telling us about local combinatorial interpretations. This is like a close cousin of what Igor Pak told us about the first day.

Pravesh Kothari is taking us on a tour through the two worlds of algorithms for tensor decomposition.

080

In the talk just before the lunch break Théo Fabris is telling us about strong separation between monotone and non-monotone circuits. There is a polynomial that has cubic size depth-3 circuit but requires exp(n) size monotone circuit.

Next, Magnus Hansen talked about the non-closure of roABPs under factoring. This is in contrast to previous results presented in the session.

Shanthanu Rai talked about the complexity of a fundamental problem, namely GCD. This is a problem I can explain to my nephew who is in 5th grade. It is fascinating that we have a new upper bound for it since 2024 and it has been made field independent in the work he presented. The talk was online.

The day started with a talk by Mrinal Kumar who presented one of my favorite recent results. The result is about the complexity of the factors of easy polynomials. It states that the factors are as easy as the polynomial. It subsumes all previous results of this kind.

Rohit Gurjar describes the result about Bipartite Perfect Matching is in NC to Avi Wigderson.

050

The paper will be put out soon. One of the authors presented almost the whole proof. The second comment above lists the authors.

110

Prahladh Harsha is telling us about how to obtain near linear time algorithms for the list deciding problem. The main bottleneck is to design a near linear time algorithm for the interpolation step that shows up in previous polynomial time algorithms.

120

Roshan Raj presented a randomized polynomial time algorithm for a natural problem called Black-box Principal Minor Assignment. A definition called rank one extensions seem very interesting. I wonder whether it has any geometric interpretation.

Computing the order of an element is an important step in integer factorization. Ben Lee Volk have an improvement to this algorithmic question.

In the afternoon session we started with a talk by Josh Alman who is telling us about Faster Walsh-Hadamard Transform from Matrix Non-Rigidity.